Getting error that I installed CPU-only version of DreamPower; I have RTX 3000-series GPU.

[arrrbeee]

Describe the bug I downloaded the latest .snap version (1.6.4) from the website and installed it with:

sudo snap install DreamTime-v1.6.4-linux-installer.snap

To Reproduce Steps to reproduce the behavior:

1. Download latest .snap file
2. Install
3. Select photo and opt to run with NVIDIA GPU (I have RTX3090)
4. Press 'Nudify'

Expected behavior I expect it to work with Nvidia GPU

Screenshots

I get the following error:

You have installed the CPU-only version of DreamPower. Please change the processing device option to CPU or reinstall DreamPower to get GPU support.

> param_applied = fn(param) File "torch/nn/modules/module.py", line 463, in <lambda> return self._apply(lambda t: t.cuda(device)) File "torch/cuda/__init__.py", line 166, in _lazy_init raise AssertionError("Torch not compiled with CUDA enabled") AssertionError: Torch not compiled with CUDA enabled [44200] Failed to execute script main
> File "processing/image.py", line 114, in _execute File "processing/worker.py", line 46, in run_worker File "transform/gan/mask.py", line 35, in __init__ File "transform/gan/mask.py", line 18, in __init__ File "transform/gan/__init__.py", line 33, in __init__ File "transform/gan/__init__.py", line 40, in __init_model File "transform/gan/model.py", line 149, in initialize File "transform/gan/model.py", line 245, in __define_g File "torch/nn/modules/module.py", line 463, in cuda return self._apply(lambda t: t.cuda(device)) File "torch/nn/modules/module.py", line 359, in _apply module._apply(fn) File "torch/nn/modules/module.py", line 359, in _apply module._apply(fn) File "torch/nn/modules/module.py", line 381, in _apply
> Traceback (most recent call last): File "main.py", line 85, in <module> File "argv/__init__.py", line 35, in run File "main.py", line 26, in main File "processing/__init__.py", line 23, in run
> [DEBUG] Loading GAN Model For correct_to_mask
> [SUCCESS] Dress To Correct Done in 0.25 seconds
> [SUCCESS] Image To Nearest Done in 0.0 seconds
> [INFO] Image resize to Nearest: 976x2800 -> 976x2800
> [INFO] Processing on ['/run/user/1000/snap.dreamtimetech/1ba39cffbd9bc777f41349cd430defbf/1675401779715/Editor.png']
> [DEBUG] To Be Executed Phases : [<class 'transform.opencv.resize.ImageToNearest'>, <class 'transform.opencv.correct.DressToCorrect'>, <class 'transform.gan.mask.CorrectToMask'>, <class 'transform.opencv.mask.MaskToMaskref'>, <class 'transform.gan.mask.MaskrefToMaskdet'>, <class 'transform.opencv.mask.MaskdetToMaskfin'>, <class 'transform.gan.mask.MaskfinToNude'>]
> [DEBUG]
> [WARNING] Image Size Requirements Unchecked.
> [DEBUG] Process to execute : <processing.image.ImageProcessing object at 0x7fd0ae7ec550>
> [INFO] GAN Processing Will Use GPU IDs: [0]
> [INFO] Welcome to DreamPower

Enviroment (please complete the following information):

• DreamTime Version 1.6.4 (latest)
• OS: Ubuntu 20.04
• RAM: 128GB
• CPU: Threadripper 2920X
• GPU: RTX 3090 24GB

Additional information

I installed PyTorch via conda using the following:

conda install pytorch torchvision torchaudio pytorch-cuda=11.6 -c pytorch -c nvidia

Per my nvcc --version output, I have CUDA 11.6 toolkit installed:

nvcc --version            (base)  Thu 02 Feb 2023 08:35:33 PM CST
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_Mar__8_18:18:20_PST_2022
Cuda compilation tools, release 11.6, V11.6.124
Build cuda_11.6.r11.6/compiler.31057947_0

When I check if cuda is available for torch with the following:

(base) ray@ray-desktop-2920x:~/Downloads$ python3
Python 3.7.7 (default, Mar 26 2020, 15:48:22) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.cuda.is_available()
True

It returns true.

I tried manually downloading the .7z archive with latest GPU+CPU version of DreamPower and extracted it to the dreampower directory per instructions. When I restart DreamTime, I see the following error:

What am I doing wrong? How do I install the GPU-version of DreamPower?